Test your physical AI on real robots and in sim.
Your coding agent submits your model. We run it and send back a video, the robot data and a score for every episode.
Trusted by
Positronic lets us evaluate checkpoints continuously as we train, so we can course-correct our research quickly. Day-to-day evaluation doesn’t compete with our robot fleet or operators for time, which keeps them focused on data collection and real-world deployment.
How it works
- 1. Sign up
Create an account in the console and add funds.
- 2. Connect your model
A container image, or an endpoint you serve.
- 3. Choose where it runs
Real robots or simulators, and the tasks.
- 4. Get the results
We run the episodes. You get a video, the robot data and a score for each one.
Robots and simulators
- Real robotFranka FR3 (DROID)from $12 per 2-min episode
- Real robotBimanual YAMfrom $12 per 2-min episode
- SimulatorRoboLabfrom $0.66 per 2-min episode
- SimulatorMolmoSpacesfrom $0.08 per 2-min episode
Running evals at scale?
For labs